1. Field of the Invention
The invention relates to a method for bidirectional data transmission between a computer system having at least two processors or a processor divided into a plurality of real or virtual subsystems with different control unit address administration and at least one peripheral unit with unique control unit address which is connected via data paths to said computers.
2. Description of the Related Art
The invention accordingly relates to a method for addressing compatibility of heterogeneous computer systems and also to an electronic apparatus for automatic control unit address conversion.
A heterogeneous computer system may consist of one or more processors, frequently of different type, and a number of peripheral units, such as disk memories, magnetic tape units, display devices and printers. In such a multiprocessor composite system for functional and economic reasons it is frequently expedient to make a peripheral device accessible to more than one processor.
To enable a peripheral unit to be addressed by a processor a unique control unit address must be assigned thereto. Hardware and/or software-induced or organisational restrictions in processors of different type may require different control unit addresses for the same peripheral unit. In such a case, the peripheral unit provided with a unique address can be used only by processors which address the peripheral unit via the same control unit address.
Hitherto it was necessary to use different peripheral units for different processors.
If a plurality of peripheral units is employed the result is a considerable financial expenditure. Even then, the computers cannot directly access the information (services) of the respective other peripheral unit. In such a case, until today for example in a magnetic tape system a human operator had to remove the magnetic tape from the magnetic tape system associated with the processor B and place it in a second magnetic tape system associated with the processor A.
Another possibility resides in switching the data paths over using a conventional channel switch but in that case the peripheral device must be reconfigured to the control unit address of the respective processor which in many systems is not possible or is not possible at any instant. It is only by employing such unsatisfactory ways that the processor A can have access to the services of the peripheral unit allocated to the processor B. This manifests itself as a considerable disadvantage in information processing, both from the financial and from the organisational point of view. This unsatisfactory condition is the state of the art.
It would be desirable to have available a method which permits free access of the processors to peripheral devices irrespective of possible hardware and/or software-induced restrictions.